Experiment
Sarbecovirus (CoV) infections, including Severe Acute Respiratory CoV (SARS-CoV) and SARS48 CoV-2, are considerable human threats. The authors used the Collaborative Cross (CC) and human GWAS datasets to elucidate host susceptibility loci that regulate CoV infections and to identify host quantitative trait loci that modulate severe CoV and pan-CoV disease outcomes including a major disease regulating loci including CCR9.
The authors infected the CC#/Unc population with two genetically distinct Sarbecoviruses, which included the clade I SARS-MA and clade II HKU3-MA strains, respectively. Groups of CC#/Unc mice were inoculated with 5x103 plaque-forming units (PFU) of SARS-MA, and viral burden, clinical disease (e.g., weight loss, mortality, and respiratory function), antibody titers, and immune cell infiltrates were measured at multiple timepoints post-infection (ranging from 2-32 days). A matched cohort of CC#/Unc mice was inoculated with 1x105 PFU of HKU3-MA and evaluated for mortality and weight loss through day 4 post infection. In both studies, virus challenge elicited an array of disease phenotypes, ranging from clinically inapparent infection to lethal outcomes within the first 4 days of infection. The authors estimated genetic contributions for many of these traits and determined that heritability for these responses was 44.4%-80.9%, estimates that agree with previous CC studies [26,28-31].
The authors conducted genetic mapping to identify both genomic regions and specific founder haplotypes driving various aspects of SARS-MA and HKU3-MA disease responses. For the CC-RIX, the authors used the same pipeline as previously described [29]. Briefly, each CC-RIX had their genome represented as an array of probabilities of each of the 8 CC founder haplotypes. This array was used in the DOQTL R package [53] to run an 8-allele regression at each of 77,000 markers for the CC-RIX phenotypes. At each marker, a LOD score is calculated describing the goodness of fit of the trait~genotype model relative to a null model. Significance was determined by running 1000 permutations scrambling the relationship between phenotypes and haplotypes. In this way, significance is independent of both population allele frequencies, as well as the phenotypic distribution. For the F2 crosses, instead of a regression on haplotype probabilities, the R/QTL package conducts a regression of the trait of interest on the exact genotypes at each locus [54]. As with the CC-RIX mapping, permutation testing is used to identify significance.
The authors identified 12 distinct and high-confidence loci in the RIX population affecting weight loss, mortality, titer, antibody responses or respiratory function after SARS-MA infection or weight loss and mortality following HKU3-MA infection (genome coordinates relative to GRCm38/mm10; haplotype information given in Table 1):
Hrsq13 (host response to SARS QTL 13, HKU3-MA mortality) maps to Chr 13: 20.3 - 23.7 Mb. The Odds Ratio (OR) for disease at Hrsq13 is 6.13.
Hrsq14 (host response to SARS QTL 14, HKU3-MA mortality) maps to Chr 5: 112.9 - 118.8 Mb. The Odds Ratio (OR) for disease at Hrsq14 is 6.13.
Hrsq15 (host response to SARS QTL 15, HKU3-MA weight loss, 4dpi) maps to Chr 15: 51.5 - 75.6 Mb. Hrsq15 contributes 3.4% of phenotypic trait variance.
Hrsq16 (host response to SARS QTL 16, SARS viral titer, 2dpi) maps to Chr 16: 18 - 51.1 Mb. Hrsq16 contributes 15.94% of phenotypic trait variance.
Hrsq17 (host response to SARS QTL 17, PenH, 2dpi) maps to Chr 11: 9.05 - 28.66 Mb. Hrsq17 contributes 11.65% of phenotypic trait variance.
Hrsq18 (host response to SARS QTL 18, Rpef, 4dpi) maps to Chr 11: 3.26 - 28 Mb. Hrsq18 contributes 6.48% of phenotypic trait variance.
Hrsq19 (host response to SARS QTL 19, IgG1 [N], 32dpi) maps to Chr 17: 27.4 - 78.3 Mb. Hrsq19 contributes 19.97% of phenotypic trait variance.
Hrsq20 (host response to SARS QTL 20, total IgG [N], 32dpi) maps to Chr 17: 27.4 - 78.3 Mb. Hrsq20 contributes 18.53% of phenotypic trait variance.
Hrsq21 (host response to SARS QTL 21, total IgG [N], 32dpi) maps to Chr 16: 18.1 - 43.8 Mb. Hrsq21 contributes 22.75% of phenotypic trait variance.
Hrsq22 (host response to SARS QTL 22, total IgG [N], 32dpi) maps to Chr 10: 17.4 - 130.5 Mb. Hrsq22 contributes 10.71% of phenotypic trait variance.
Hrsq23 (host response to SARS QTL 23, total IgG [N], 32dpi) maps to Chr 3: 3.1 - 35.4 Mb. Hrsq23 contributes 10.83% of phenotypic trait variance.
Hrsq24 (host response to SARS QTL 24, CD8+ DCs [%], 4dpi) maps to Chr 6: 67.2 - 85 Mb. Hrsq24 contributes 8.17% of phenotypic trait variance.
Concurrent with the large CC#/Unc screen, the authors identified a pair of inbred CC strains showing highly divergent susceptibilities to SARS-CoV: the disease resistant CC011/Unc and the highly susceptible CC074/Unc strain. After challenge with 1x104 PFU SARS-MA, these strains exhibited marked differences in clinical and virological disease phenotypes (e.g., hemorrhage, weight loss, virus titer, mortality, circulating immune cells), and all CC074 mice developed lethal disease by 4 dpi post infection. Relevant for the current pandemic, these parental strains showed similar severe infection phenotypes during SARS-CoV-2 MA10 infection.
The authors generated 403 F2 mice by intercrossing these strains and inoculated them intranasally at 9-12 weeks with 1x104 161 PFU of SARS-MA. These F2 mice showed an expanded range of disease responses relative to their parent CC strains, including mortality, weight loss, titer, respiratory function, circulating immune cell and hemorrhage phenotypes. CC003, CC053, their F1 progeny, and the F2 cross were genotyped as previously described [45]. CC011, CC074, their F1 progeny, and the F2 cross were genotyped on the MiniMUGA genotyping array [52]. The authors conducted QTL mapping in these F2 mice and identified 10 significant QTL segregating in this population (CC011 x CC074) F2:
Hrsq25 (host response to SARS QTL 25, mortality) maps to Chr 4: 32.95 - 114.54 Mb. The Odds Ratio (OR) for disease at Hrsq25 is 4.34.
Hrsq26 (host response to SARS QTL 26, weight loss, males, 4dpi) maps to Chr 4: 6.38 - 17.97 Mb. Hrsq26 contributes 12.57% of phenotypic trait variance.
Hrsq27 (host response to SARS QTL 27, mortality) maps to Chr 9: 74.94 - 124.06 Mb. The Odds Ratio (OR) for disease at Hrsq27 is 3.15.
Hrsq28 (host response to SARS QTL 28, hemorrhage, 4dpi) maps to Chr 9: 117.38 - 124.07 Mb. Hrsq28 contributes 10.24% of phenotypic trait variance.
Hrsq29 (host response to SARS QTL 29, PenH, 2dpi) maps to Chr 9: 116.24 - 124.07 Mb. Hrsq29 contributes 7.76% of phenotypic trait variance.
Hrsq30 (host response to SARS QTL 30, periph. neutrophils, 4dpi) maps to Chr 9: 111.54 - 122.63 Mb. Hrsq30 contributes 11.8% of phenotypic trait variance.
Hrsq31 (host response to SARS QTL 31, periph. lymphocytes, 4dpi) maps to Chr 9: 111.54 - 122.63 Mb. Hrsq31 contributes 12.39% of phenotypic trait variance.
Hrsq32 (host response to SARS QTL 32, periph. neutrophils, 4dpi) maps to Chr 11: 26.44 - 80.76 Mb. Hrsq32 contributes 5.5% of phenotypic trait variance.
Hrsq33 (host response to SARS QTL 33, periph. lymphocytes, 4dpi) maps to Chr 11: 17.89 - 80.76 Mb. Hrsq33 contributes 5.6% of phenotypic trait variance.
Hrsq34 (host response to SARS QTL 34, PenH, 2dpi) maps to Chr 15: 58.66 - 74.04 Mb. Hrsq34 contributes 8.48% of phenotypic trait variance.
CCR9 emerged as a strong candidate based on the integration of the data with these studies and the presence of nonsynonomous SNPs in CCR9 as well as synonymous mutations in regulatory flanking sequences.
Next, the authors revisited a previous CC-F2 intercross, CC003/UncxCC053/Unc (named CC003 and CC053 from here on) conducted by their group [45], and utilized their refined analysis pipelines once the original SARS-MA disease loci (HrS5-9) were statistically accounted for. This re-analysis allowed them to identify two additional loci, as well as several other suggestive QTL:
Hrsq35 (host response to SARS QTL 35, hemorrhage, 4dpi) maps to Chr 4: 35.61 - 104.04 Mb. Hrsq35 contributes 6.86% of phenotypic trait variance.
Hrsq36 (host response to SARS QTL 36, weight loss, 4dpi) maps to Chr 4: 35.61 - 104.04 Mb. Hrsq36 contributes 6.22% of phenotypic trait variance.
Hrsq35 and Hrsq36 also overlapped with the mortality QTL Hrsq25 identified in (CC011 x CC074) F2 cross. Variation between CC003 and CC053, as well as between CC011 and CC074 in this locus pointed to Trim14 as a likely candidate gene driving these differences in SARS-CoV disease.
The authors used CRISPR/Cas9 targeting to edit Trim14 in C57BL/6J mice, create a functional knockout, and evaluate its role following infection. Trim14Delta47/Delta47 mice inoculated with 1x105 PFU of SARS-MA had a modest increase in pathogenesis relative to C57BL/6J control mice. At 3 and 4 dpi, Trim14-deficient mice had increased weight loss, which corresponded with increases in viral titer within the lung at 2 and 4 dpi. This result show that an absence of Trim14 affects viral clearance.
Similarly, Trim14Delta47/Delta47 mice inoculated with SARS-CoV-2 MA10 also sustained modest increases in weight loss and a delayed recovery phenotype when compared to C57BL/6 mice. However, the difference in viral titer seen at early times post SARS-MA infection was not observed with SARS-CoV-2 MA10. Together, these data suggest that Trim14 has a shared role in attenuating Sarbecovirus disease potential, but that this effect varies between viruses.
Analysis Tools